I have been strugling with errors in stl files generated in v8. I’ve downloaded several third party repair programs, but it takes multiple passes using Rhino Mesh Repair, MeshLab, and FreeCAD to reduce the errors so I can get a good print - I’m working with very large and complex structures. Unfortunately Shrinkwrap reduces the sharpness of small details, making it less than ideal. I just tried the latest WIP on a structure that has been giving me fits and the slicer easily fixed the few errors that were still there. It looks like I’ll have to upgrade to v9. Thanks.
You still send STL files to your slicer instead of STEP?
1.) Some slicers still don’t support .STEP
2.) Better tessellation control exporting via cad. Not a big issue but for certain parts it might be warranted - especially detailed resin prints.
@hackettet Your original geometry might be problematic?
I’ve experimented with most of the common slicers, Prusa, Orca, Chitubox, Satellite, and Lychee. Chitubox is more organized than the others, but none of the auto support functions are useful. I have to manually apply all the supports. Lychee is the only one that has the flexibility and user interface I need, but it accepts only stl files. Combining all the components into a single Boolean union (small files have ~400,000 surfaces, some are close to 1 M) greatly reduced the error rate, but there are still hundreds of problems, mostly naked edges, non manifold edges, overlapping surfaces, and holes. Lychee can fix holees but can’t handle the other problems. SateLite fixes everything but in the process strips away the fine detail I need. Between MeshLab and FreeCAD I can fix most of them. Unfortunately, I haven’t been able to install AutoCAD Meshmixer and after years of trying, I still can’t figure out how to use Blender.. By carefully inspecting the slicesI can determine how well Lychee will print the file even with some errors still in place and it’s usually OK.. The stl from the WIP version had only a few holes, which Lychee fixed on opening.
Then don’t use those pieces of junk, it’s 2026.
It’s a whole lot easier to keep to my workflow and use STL files than go about changing slicers in a business with multiple people for literally no reason at all. There’s also no real benefit to STEP - you’re still exporting a file at the end of the day. And I prefer defining the mesh quality during cad export. Especially for medical parts with tight tolerances - the tessellation in slicer is worse than what you get in cad with high quality mesh settings.
Personally I have rarely seen crappy Rhino mesh exports from good, closed, valid, manifold Rhino polysurface objects. It would be good to have an example of one of those if you can provide one, then at least they have a clue as to how the mesh engine can be improved. Also be good to see an example were the WIP meshes things better (or worse) than V8.
Me too. At least that way I know what I am going to get. But yes, I am old-school.
What slicer are you using?
Sect B Stringers.3dm
Looks like I started a firestorm! Sorry about that. I’m using Lychee slicer because it can stick “mini supports” into very small places. Attached is a small component of the model I’m building of the bridge across the New River at Thurmond, WV in the New River National lPark. It’s a boolean union of about 1030 joined (solid) surfaces and represents about 2% of the total structure. After a quick look, it appears that Prusa Slicer is the only one that supports STEP files. Filament printers are useless at the level of precision I require.. I need precision to at least .001" (0.025mm) to ensure components fit together. I design a clearance of .002" for adhesive - Mil Specs often require .005" depending on the type of adhesive and substrate, but they also use fixtures to hold parts in position… Filament printers don’t even come close and many resin printers struggle to attain that level of accuracy. .My Saturn 4 Ultra isn’t perfect, but it takes onay a small amount of post printing work to get difficult parts to fit.
It happens sometimes a firestorm and pitchforks 
Interesting slicer first time I’ve heard of it
Well, there are a number of problems with the file, primarily starting with the file absolute tolerance. You might have a read here.
Before that, after ungrouping everything, you have 441 duplicate objects in there that are selected by SelDup. In addition, on the Tie Indexes layer you have 4 nearly identical objects that SelDup doesn’t select. Including all those in your mesh will only cause problems.
Back to tolerances, your file tolerance of 0.01 is far too loose for the level of detail you are modeling, some details are 0.1mm or smaller, which approaches the absolute tolerance too closely. I would model this stuff with 0.001 tolerance minumum, and maybe even 0.0001 (more than that won’t help).
Unfortunately, having modeled this stuff with those too-loose file tolerances, forcing things together with BooleanUnions and the like will make bad edges, non-manifold objects etc. which will be transmitted in the mesh. After changing the file tolerance to 0.001, I tried a number of fixes which involved moving some things tiny amounts to get them to line up better. Unfortunately, the edges of most parts have been modeled with looser tolerances than that, so I can get closed solid polysurfaces, but running Explode/RebuildEdges/Join makes everything pull apart again.
After quite a bit of work, I did manage to get things back together going back to 0.01 tolerance and end up with one valid, closed, nonmanifold plysurface. I then meshed it, with a couple of different parameter tries. The object is to get a mesh with enough detail to cover your 1000 rivet heads at 0.4mm in diameter, but not more than is necessary for your 3D printer.
The above resulted in a closed mesh with 8m+ polygons, and the .stl is over 600Mb.
I tried a “lighter” version which made only 3m+ polygons, ~220Mb file - this onle looks largely sufficient IMO…
I even went a little lighter to make a 1.3m polygon mesh with the following settings, a few more detail artifacts, but for 3D printing may still be more than adequate. Only 85Mb.
I don’t have any slicers here other than the one for my Bambu Lab P1S, so I can’t tell how good the .stls are - I’ll send you the fixed Rhino file and let you mesh it yourself and report back.